In the last decade, disinformation has become a matter of widespread concern among governments, citizens, international bodies, digital platforms and academics. But behind this concern lies a key question: What’s the magnitude of the phenomenon?

Answering this question is immensely complex. Disinformation is spread in multiple media, formats and platforms: from social media and websites to messaging apps, traditional media outlets and daily conversations. This diversity makes measuring it an enormous challenge. However, we can employ some methods to try estimating it, although each comes with important limitations.

Analysis of URLs and Website Quality

---

The first approach — the most commonly used when analyzing large volumes of data on social media — focuses on examining links to external websites that often accompany certain posts, and uses the quality of the destination website as an indicator of the truthfulness of the content. In principle, links to websites with lower information quality are more likely to be false or inaccurate.

To identify if a post includes a link to a disinformation website, these studies compare URLs with lists of disinformation sites compiled by experts. However, these lists have their constraints. They exclude many websites that are not news outlets, while others are left out for being new or relatively unknown. Even though these lists have expanded in the last few years, both in quantity and inclusion criteria (for example, websites that are not news outlets), they remain limited. Besides, many URLs lead to other platforms, where the problem is not the site itself, but its specific content.

This method presents other significant limitations. By focusing solely on posts with external links, it excludes texts, images, videos and claims that circulate without links to external sites. For practical reasons, it also often works with data from Twitter (now X), but this platform is only used by a small part of citizens.

Other highly relevant digital channels are disregarded, like WhatsApp or Telegram, where information spreads privately. It also doesn’t take into account claims made by political leaders, spokespeople or influencers that are widely trusted and have a massive reach. Finally, it excludes all disinformation disseminated through traditional mediaoutlets — radio, television, newspapers — and offline spheres like daily conversations or community spaces.

Surveys and Experiments

---

The second approach consists of gathering headlines, posts or specific articles that have been debunked or are true and asking participants how likely they would share them if they see them online.

This method removes potential exposure bias, where some users are more exposed to disinformation than others. However, it also has its limitations: It is difficult to implement on a large scale and it is based on declared intentions rather than actual sharing behavior on social media.

New Methodologies

---

In the last few years, studies started to use methods of combined analysis to overcome these limitations. For example, instead of comparing URLs with only one list, they are compared with multiple lists of disinformation sites or index of information quality created by different experts and with wider and more varied criteria. Another approach to assessing the quality of news or websites combines the criteria of experts with evaluations politically balanced groups of citizens, thus trying to find more robust answers to the problem of measuring the amount of disinformation circulating in society.

In summary, although it is impossible to know all disinformation that is being spread, we can use different methods according to our object of study. What’s important is to clearly know what slice of reality we’re taking and the limitations of our approaches. All scientific and methodological research have inherent restrictions, which doesn’t constitute a flaw in itself, but an intrinsic feature of the academic method. What’s important is not the absence of these limitations — which is impossible — but to be aware of them and explicitly state them. Only then can we correctly understand the reach and consequences of our conclusions and avoid improper generalizations or interpretations that exceed what data truthfully allows us to ascertain.